Cause of death: where there is no physician
Overview of verbal autopsy
VA can be used to determine COD in settings where deaths occur such that there is no medical certification of cause of death (MCCOD). This may be outside of hospitals or in health facilities with only limited diagnostic capabilities. The aim of VA is to generate estimates of COD distributions (cause-specific mortality fractions, CSMFs) at the population level.
In VA, the diagnosis of causes of death is based on responses that a health worker collects from families and caregivers who are asked a series of questions about the nature and duration of the signs and symptoms that the deceased experienced before death. The responses are analysed to determine the probable COD. There are two approaches to this:
- One or more physicians review the responses and reach a conclusion on the most likely COD (referred to as physician-certified verbal autopsy or PCVA)
- Automated methods are applied to come up with a probable COD.
The fundamental and most important policy-relevant outputs of VA are the numbers of deaths in the population due to various causes, as fractions of the total. These are known as cause-specific mortality fractions (CSMFs).
Because VA is mostly conducted in settings where there is no physician, in practice PCVA requires that the completed interviews be sent out to available physicians for review, or collected until a local physician has the time to review them. PCVA can thus be time consuming and burdensome for physicians who are generally required to prioritise their clinical responsibilities over other activities. By contrast, automated methods can be completed quickly, at little expense, immediately after the VA interview.
VA using PCVA has been used for many years to generate COD data in health and demographic surveillance sites, many of which carry out research on the effectiveness of interventions designed to improve health. Because these sites are not selected to be nationally representative, the data they generate on causes of death are mainly useful at the local level. There are a few examples of the use of VA in nationally representative sample sites (eg in India) but these are stand-alone activities not yet linked to the CRVS system. However, the skills, knowledge and experience gained through application of VA in these sample or demographic surveillance sites could and should be used to support the routine application of VA in CRVS systems. This will be addressed later in this section.
De Savigny D et al (2017). Integrating community-based verbal autopsy into civil registration and vital statistics (CRVS): system-level considerations. Global Health Action.